Last week I visited VMworld 2014, where a horde of steely-eyed virtualization acolytes swarmed into the din of an expo floor, creating an atmosphere redolent of a religious revival being held in a Vegas casino. Vendor booths filled Moscone Center's South Hall to bursting, ranging from enormous displays presented by longtime technology stalwarts such as HP and IBM down to a multitude of tiny booths displaying products from new market entrants. All came to pay fealty to the industry's dominant virtualization player and claim their place in the VMware firmament. The outpouring of energy (and money) made the show floor something like a natural wonder akin to the Niagara Falls – overwhelming, awe-inspiring and vastly entertaining.
[ Feature: Hot Products at VMworld 2014 ]
However, in terms of learning, I found much more food for thought at a 451 Group event I attended, where Peter Ffoulkes presented a summary of the results from the firm's 451 Research Cloud Computing Wave 7 study, along with commentary discussing the implications of the survey. It's fair to say that the survey pool represents the apotheosis of enterprise IT – large, mainstream companies with significant internal IT organizations.
My ears perked up when, early in his presentation, Ffoulkes noted, "Enterprises aren't using AWS." As you can see from the slide, his statement aligns with the survey results. In two years, only 18 percent of all workloads will be in public cloud environments, and less than half of that will be non-SaaS.
The message is clear: The vast majority of workloads will remain on premises for the foreseeable future, with only a tiny percentage running in AWS. However, I think the comforting inference many vendors and IT groups draw from this particular result is misplaced.
As far as cloud computing itself goes, it's early. As another slide from the presentation indicates, only 17 percent of enterprises have a private cloud in place; barely a quarter have advanced as far as automation, with a majority still applying fully manual methods to a virtualized infrastructure. Ffoulkes' observation: It would take most IT groups two refresh cycles to get to a true on-premises cloud computing infrastructure.
Certainly, the show floor displays confirmed the 451 survey results – endless displays of storage and network kit, along with many virtualization add-on products designed to operate VMware products more efficiently. In a phrase, it was system administrator heaven.
While the phrase cloud computing was splashed across booth after booth, you couldn't help but feel that the vast preponderance of the energy in the exhibition hall was devoted to extending the current virtualized environment, with an emphasis on incremental improvement, and certainly not a significant disruption to business as usual.
As far as addressing the "developer" part of DevOps – that is, the user of virtualized resources – it's safe to say that developers represented a shadowy presence hovering beyond the shadows of the campfire around which VMworld attendees huddled. Unmistakably, the needs of developers and application groups were viewed as a kind of generalized, inchoate demand to be addressed at some point, once the infrastructure was all shipshape.
Understanding the Nuances of AWS Spending
It's here that the comforting inferences many might draw from the survey show their true danger. When I queried another analyst on the conclusion that enterprises aren't using AWS, he restated the survey results in a more nuanced way. He noted that enterprises are using AWS, but only for a subset of applications – collaboration, some analytics and Web properties, to name three. The vast majority of enterprise applications still run in corporate data centers. Put differently: While there is enterprise spending with AWS, it's a tiny fraction of overall spend and, therefore, relatively unimportant.
While I respect the analyst, I respectfully disagree. Too many people look at the ratio of total spend to public cloud spend and conclude that public cloud computing is a rounding error. That fails to understand the rate of change (their first derivatives) in the two areas. Even more importantly, it fails to comprehend their second derivatives, which is how fast they're changing. As IDC wrote in its 2014 forecast, the IT vendor space is really two markets. There's a stagnant (0.7 percent growth) second platform – a "legacy IT" market – and a high-growth (15 percent) third platform market.
That floor full of vendors offering network, storage and VMware extensions? It's a collection of legacy-focused companies locked in a zero-sum game, where one firm's growth comes at the expense of another vendor. One can certainly see this in big tech's financial results:
- IBM revenue has been shrinking for seven straight quarters.
- Though HP announced its first revenue rise in three years, the company is laying off 16,000 employees.
- Cisco Systems is letting 8 percent of its workforce go.
Every new vendor that crowds into the virtualization ecosystem makes its bones by stealing revenue from an incumbent.
On the public cloud side? As I wrote last month, Pacific Crest Securities estimates that AWS will grow 40 percent per year for the next five years. This significantly outstrips IDC's estimate. I believe Pacific Crest's estimate more nearly represents the third Platform growth rate. In my experience, the growth rate is increasing; that is, its second derivative is growing.
[ Commentary: VMworld Promises an Amazing Future of Virtual Machines ]
I find the whole message of VMworld puzzling. Attendees could come away from the event confident of a measured pathway to the future, secure in the knowledge that an entire ecosystem of vendors will make it possible for them to incrementally improve their existing environments.There seems to be little awareness that this approach won't nearly satisfy the increased expectations of developers. Now that they've seen what's possible from AWS (or Google or Microsoft, for that matter), the belief that they'll wait two refresh cycles for functionality approximating what's widely available today is risible.
All too often, developers meeting their own needs by using public cloud computing gets characterized as shadow IT or rogue IT, as though labelling the phenomenon with a pejorative term will delegitimize it or, even better, make it stop. As both IDC and Pacific Crest indicate, this phenomenon, if anything, is accelerating.
What becomes clearer each day is that two groups cohabit the world of IT, each with a perspective on what's required to address the IT needs of the business world. All too often, unfortunately, mutual miscomprehension or even contempt separates the two groups. While it's clear who holds the clout today, the drastically different growth rates represented by the second and third platforms auger whom the ultimate victor will be.